Astropulse v6 has been released for linux and windows. |
![]() |
| log in |
Message boards : News : Astropulse v6 has been released for linux and windows.
1 · 2 · 3 · Next
| Author | Message |
|---|---|
|
We've released the Astropulse version 6 application to the SETI@home project. The only real difference is a couple of minus signs in the code that were added to match changes to the way the data recorder works. | |
| ID: 1205550 · | |
|
Thanks for the "Heads-up" Eric. | |
| ID: 1205570 · | |
|
Questions and problems to Updating for upcoming apps..... in the Number Crunching forum, please. | |
| ID: 1205573 · | |
|
Is Astropulse V6 going to ever be released for OSX? Or is that just another unsupported feature of SETI BOINC on OSX? I am considering stopping donations to SETI until my chosen platform is completely supported!!! | |
| ID: 1205574 · | |
It'll take a while for the credit granted by this application to settle out to a reasonably constant value. ... "a while" as in years? For the "old" apps the credit system still seems to be a random number generator ... | |
| ID: 1205802 · | |
|
Preaching to the choir on that one. The FLOP counting credit system was far more stable than the current one. | |
| ID: 1205808 · | |
We've released the Astropulse version 6 application to the SETI@home project. The only real difference is a couple of minus signs in the code that were added to match changes to the way the data recorder works. Dear Eric, Thank you very much for this update. Keep searching Best wishes Byron | |
| ID: 1205822 · | |
|
Is AstroPulse going to support CUDA? It could really use the GFLOPS. =) | |
| ID: 1206020 · | |
|
We plan that eventually it will support CUDA and OpenCL. Unfortunately Astropulse has lots of twisty turny passages, so I don't expect the speedup will be that huge. It'll probably spend a lot of time using CPU to avoid bogging the GPU down with unproductive stuff. | |
| ID: 1206218 · | |
|
"twisty turny passages" huh? Is that technical phrasing? ;) | |
| ID: 1206230 · | |
|
"grad student code" seemed too mean. ;) | |
| ID: 1206235 · | |
"grad student code" seemed too mean. ;) :) ____________ | |
| ID: 1206285 · | |
We plan that eventually it will support CUDA and OpenCL. Unfortunately Astropulse has lots of twisty turny passages, so I don't expect the speedup will be that huge. It'll probably spend a lot of time using CPU to avoid bogging the GPU down with unproductive stuff. Thanks Eric. I know you have a lot to do so I do thank you for the time spent. ____________ | |
| ID: 1206310 · | |
|
Based on your post I changed my Preferences to "No" for Astropulse v6 but then shortly thereafter received some downloads. I assume I received them because I had indicated an openness to receive "Other applications" if applications I agreed to were not available. Since I don't run "optimised" I'll do them, and I heve closed the loophole (if that is what happened. | |
| ID: 1206697 · | |
|
Yeah, that's always been the problem with Astropulse, since it can take days for a result to be computed, and a computer might not start computing it for a week or more sometimes you'll wait weeks for credit. But in your "Recent Average Credit" it should even out eventually. | |
| ID: 1206724 · | |
|
Just got 832.47 credits for an astropulse unit which took 1350353.04 s on my Solaris 11 Virtual Machine. The BOINC client and the app are by Dotsch. My CPU is an Opteron 1210 at 1.8 GHz running SuSE Linux 11.1 32-bit. But the Solaris 11 guest OS is a 64-bit. | |
| ID: 1206873 · | |
Just got 832.47 credits for an astropulse unit which took 1350353.04 s on my Solaris 11 Virtual Machine. The BOINC client and the app are by Dotsch. My CPU is an Opteron 1210 at 1.8 GHz running SuSE Linux 11.1 32-bit. But the Solaris 11 guest OS is a 64-bit. Congratulations on completing such an epic task with just 18 hours to spare. But note that your app is the older v505 type. You will have to ask Dotsch to compile the new v6.01 app for Solaris if you wish to continue the quest. | |
| ID: 1206883 · | |
|
I've been getting 1-2 Astropulse projects per week for quite a few months. Nothing out of the ordinary about them. Suddenly a couple of days ago I got 2 HUGE astropulse projects that were easily 3-4 times the size of any of the previous ones & one was labled "High Priority". What made it high priority? | |
| ID: 1207354 · | |
|
Those two are of the new version (v6.01). I assume by "HUGE astropulse projects" you mean "AstroPulse tasks with unusually long runtime estimates" ;-) | |
| ID: 1207382 · | |
I've been getting 1-2 Astropulse projects per week for quite a few months. Nothing out of the ordinary about them. Suddenly a couple of days ago I got 2 HUGE astropulse projects that were easily 3-4 times the size of any of the previous ones & one was labled "High Priority". What made it high priority? Questions like this are usually best asked in the Number Crunching or Q & A sections of this message board. But the point you've raised is a general one, and no doubt others will be wondering the same thing - so let's deal with it here. In actual practice, you will find that the new tasks run for very much the same length of time as the previous ones. Previously, you would occasionally get a task which finished abnormally quickly: there should be fewer of those, this time. The 'huge' task size you're seeing is actually just an estimate - and a bad estimate, at that. BOINC is notoriously bad at making initial estimates when something new comes along: even though this change is relatively small, BOINC throws away all its accumulated knowledge of how fast your machine is, how big the tasks really are, and so on: it starts again from scratch. And, starting from scratch, it's thinking (initially) that you may have difficulty completing the task before the deadline set for its return. That's all that 'high priority' means - 'do this job before any of the others, just in case'. Once it has been running for a few hours, the estimate will become more realistic and that 'high priority' flag should disappear. That's for this task. It takes some time to persuade BOINC (again) to accept how fast your machine really is. The next Astropulse task you receive will have the same inflated runtime estimate as this one. And the next. And the one after that. You will, I'm afraid, have to complete at least ten Astropulse tasks - possibly more, depending on the luck of the draw - before things return to normal. In the meantime, just ignore these queue-jumpers and let your machine go about its business as usual. | |
| ID: 1207384 · | |
Message boards : News : Astropulse v6 has been released for linux and windows.
| Copyright © 2013 University of California |